| Deconvolution procedures...Anybody have any? [message #269] |
Tue, 10 March 1992 09:53  |
howp
Messages: 3
Registered: March 1992
|
Junior Member |
|
|
Hello. I am using IDL to analyze data for my thesis. Unfortunately, I
have not found a general deconvolution procedures. I have used an
inverse FFT to perform this, but from what little I understand an inverse
FFT assumes that the array being transformed is periodic, and my data is
not necessarily so. I was wondering DOES ANYBODY HAVE A GENERAL DECONVOLUTION
PROCEDURE? Maybe IDL already has one, but I haven't found it.
Thanks.
Peter How (The Perpetual Grad. Stud.)
Institute of Space and Atmospheric Studies
University Of Saskatchewan
Saskatoon, SK
S7N 0W0
************************************************************ ***************
* Wit is... brevity *
************************************************************ ***************
|
|
|
|
| Re: Deconvolution [message #8714 is a reply to message #269] |
Thu, 17 April 1997 00:00  |
Achim Hein
Messages: 42
Registered: February 1996
|
Member |
|
|
Eddie H. Snell wrote:
>
> I've been trying to write (unsuccessfully) a simple routine to
> deconvolute a known signal from a measured stream of data giving
> a data set without the effects of the known signal.
>
> The method I have been using is a simple fft of the measured
> data and known signal, dividing the fft's by each other then
> doing the reverse fft. This doesn't work too happily (it looks
> like ailiasing may be taking place and I need a filter). I
> have looked through the IDL libraries for simple deconvolution
> module but have not found one. has anyone tackled this problem
> before and come up with a working solution :)
>
If your signal is known, why don't you look for the one ore some more
maximum peaks (depends on) while correlating the measured stream with
the known signal. If there is the searched signal in your measured
stream you will get a kind of peak if you correlate the noisy stream
with the known signal structure. That's called auto-correlation.
You can do this in the frequency domain by multiplying stream with the
conjugate complex of the known signal...
You are trying to deconvolute the signal by a kind of inversefiltering
in the frequency domain, so beware of zero padding to avoid additional
aliasing effects.
I think in principal the way you do it should work (zeropadding) but
generally filtering noisy signals with inverse structures is quite
simple but not very succesful way of filtering - unless you can reduce
the noise.
Cheers
Achim Hein
--
_______________________________________________
Dipl.-Ing. A. Hein
PB2 / ZESS - Uni-GH-Siegen
Paul-Bonatz Str. 9-11
57068 Siegen
Phone: 0271/740-3362
Fax: 0271/740-2336
Mail: Hein@nv.et-inf.uni-siegen.de
_______________________________________________
Please have a look at our Web-Sites:
http://www.nv.et-inf.uni-siegen.de/pb2/www_pb2
_______________________________________________
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
